Sei sulla pagina 1di 5

File name _Flight Delay

Classifier: Naive Bayes

Confusion Matrix

y_pred
0 1
0 443 0
1 97 10

Finding out Accuracy, precision, Sensitivity and Specificity

# Accuracy= (TP+TN)/(TP+TN+FP+FN)

#Precision= (TP)/(TP+FP)

#Sensitivity= TP/(TP+FN)

#Specificity=TN/(TN+FP)

Accuracy 82.36
Precision 100

Sensitivity 9.345

Specificity 100

Classifier : KNN Classification

y_pred
0 1
0 431 12
1 100 7

Accuracy 79.63636

Precision 36.84

Sensitivity 6.542056

Specificity 97.29

Classifier :SVM Classification

Confusion Matrix
y_pred
0 1
0 443 0
1 97 10
Finding out Accuracy , precision , Sensitivity and Specificity

Accuracy 83.27
Precision 100

Sensitivity 9.345

Specificity 100

Classifier : Decision Trees Classification

y_pred
0 1
0 443 0
1 97 10

Accuracy 83.27
Precision 100

Sensitivity 9.345

Specificity 100
Classifier : Random Forest Method Classification

Confusion Matrix

y_pred
0 1
0 430 13
1 91 16

Finding out Accuracy , precision , Sensitivity and Specificity

Accuracy 81.09
Precision 55.17

Sensitivity 14.95

Specificity 97.06
Name Accuracy Precision Sensitivity Specificity

Naives Bayes 82.36 100 9.345 100

KNN 79.636 36.84 6.54 97.29

SVM 83.27 100 9.35 100

Decision Tress 83.27 100 9.35 100

Random Forest 81.09 55.17 14.95 97.06

Interpretation
1- Accuracy was mostly same for all the classification models, but the major difference
was the sensitivity of different models. Random forest came out to be the best
classification model with a sensitivity of 14.95 for this data set with a split of 0.75.
2- For KNN Classification, the number of neighbor’s taken were 71, so the accuracy was
lowest amongst the all models, a greater number of neighbors can increase the
accuracy of the models.
3- Random Forest method provided Powerful and accurate result, a better model with
respect to others, only problem that we can face is overfitting, that can easily occur
and a need to choose the number of Trees.

Potrebbero piacerti anche